Bard/Magi/Palm2: serving infrastructure and efficiency optimization
GLaM: Efficient Scaling of Language Models with Mixture-of-Experts
GSPMD: General and Scalable Parallelization for ML Computation Graphs
GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding
GPipe: Efficient training of giant neural networks; Automatically discovery of deep learning architectures using meta learning.

AmoebaNet: Regularized Evolution for Image Classifier Architecture Search
Alpa: Automated Model-Parallel Deep Learning
Deeprank: The Algorithm That Made Search Natural. https://youtu.be/tFq6Q_muwG0?t=2631
M4: Massively multilingual massive neural machine translation

I'm heavily involved in these projects above